[Paper Review] Optimal Equilibria of the Best Shot Game

Luca Dall’Asta, Paolo Pin+4 more|arXiv (Cornell University)|Jan 1, 2009

Game Theory and Applications10 references4 citations

TL;DR

This paper proposes a simulated annealing-based mechanism that, through iterative best-response updates, converges to an optimal equilibrium in the best shot game—where the fewest nodes play action 1—on arbitrary networks, even without prior knowledge of the network or equilibrium structure. The key contribution is a provably convergent, implementable process that minimizes costly public good provision in networked public goods games.

ABSTRACT

We consider any network environment in which the "best shot game" is played. This is the case where the possible actions are only two for every node (0 and 1), and the best response for a node is 1 if and only if all her neighbors play 0. A natural application of the model is one in which the action 1 is the purchase of a good, which is locally a public good, in the sense that it will be available also to neighbors. This game typically exhibits a great multiplicity of equilibria. Imagine a social planner whose scope is to find an optimal equilibrium, i.e. one in which the number of nodes playing 1 is minimal. To find such an equilibrium is a very hard task for any non-trivial network architecture. We propose an implementable mechanism that, in the limit of infinite time, reaches an optimal equilibrium, even if this equilibrium and even the network structure is unknown to the social planner.

Motivation & Objective

• To identify and reach an optimal Nash equilibrium in the best shot game, defined as one with the minimal number of nodes choosing the costly action (1).
• To design a mechanism that achieves this optimal equilibrium without prior knowledge of the network structure or the target equilibrium.
• To ensure convergence to such an optimal equilibrium through a decentralized, implementable process based on best-response dynamics.
• To formalize the convergence of best-response dynamics across different equilibria, proving that any two equilibria are connected via a finite sequence of single-node action flips.

Proposed method

• The mechanism uses a stochastic best-response update rule (F) that selects an unsatisfied node uniformly at random and flips its action to 1, triggering a deterministic best-response propagation (B) among neighbors.
• The propagation of best response is limited to the second neighborhood of the flipped node (N1_i ∪ N2_i), ensuring finite convergence time.
• The process alternates between stochastic action flips (F) and deterministic best-response updates (B), simulating a form of simulated annealing.
• Convergence to a new Nash equilibrium is guaranteed in finite steps due to the finite size of the network and the structure of maximal independent sets.
• The mechanism is proven to converge to any target equilibrium from any initial state through a finite sequence of single-node action changes.
• The method relies on the mathematical property that any two maximal independent sets (i.e., equilibria) are connected via a finite path of single-node flips, each followed by full best-response propagation.

Experimental results

Research questions

• RQ1 Can a decentralized mechanism converge to an optimal equilibrium in the best shot game, defined as the equilibrium with the fewest players choosing action 1, without prior knowledge of the network or equilibrium structure?
• RQ2 What is the structure of the state space of equilibria in the best shot game, and how are different equilibria connected via best-response dynamics?
• RQ3 Is there a finite, implementable process that can reach a socially optimal equilibrium (minimal cost) from any initial configuration in a networked public goods game?
• RQ4 How can best-response dynamics be structured to ensure convergence to a desired equilibrium, even when the target is unknown?
• RQ5 What are the bounds on the number of steps required for convergence from any initial state to a new equilibrium via best-response updates?

Key findings

• The mechanism converges to an optimal equilibrium in finite time, even when the network structure and the target equilibrium are unknown.
• Best-response propagation is bounded to the second neighborhood of any flipped node, ensuring finite convergence time.
• Any two equilibria (i.e., maximal independent sets) are connected via a finite sequence of single-node action flips, each followed by full best-response propagation.
• The process guarantees convergence to a new Nash equilibrium after each action flip, due to the finite and discrete nature of the state space.
• The method ensures that each intermediate state remains a valid equilibrium after every action flip and propagation step.
• The convergence path is deterministic after each stochastic flip, and the entire process is provably convergent to a new equilibrium, potentially optimal.